Systematic Review of Methods for Individual Patient Data Meta-Analysis with Binary Outcomes

• Activity Number: 550
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Andrea Benedetti*+ and Doneal Thomas
• Companies: McGill University
• Keywords: individual patient data ; meta analysis ; random effects ; systematic review ; heterogeneity

Abstract

Meta-analyses based on individual patient data (IPD-MA) are regarded as the gold standard and are becoming increasingly common. In a review of articles published from 1999-2001, the statistical approach was seldom reported in sufficient detail, and the outcome was binary in 32% of the studies considered. Here, we review 26 IPD-MA published in 2011 that used a binary outcome to consider: the approach used in their analysis, (e.g. a one- or two-stage model), the assumptions of fixed or random effects, etc. We investigated how heterogeneity across studies was described and methods used to adjust for covariate effects. 18 of 26 IPD-MA used a one-stage approach. Random-effect logistic regression was the standard method, allowing the treatment effect to vary across studies. Most IPD-MA presented some form of statistic to measure heterogeneity across studies, usually calculated using a two-stage approach. Subgroup analyses on patient-level covariates was the general approach of analyzing the covariate- treatment interaction effects. IPD-MA with binary outcomes has increased. Methods are still often not reported in enough detail.